Hey Fintech Nerds 👋,
Zurich was super interesting. European regulators are suddenly much more open minded about stablecoins, almost a full 180 from a year ago when GENIUS passed. The rise of bank stablecoin initiatives like Qivalis have really helped here.
I’m en route back to London, and there’s so much news. Prediction markets keep being drama, from the CFTC suing Kentucky, to alleged misleading ads by Polymarket, and even Meta getting in on the act. Airwallex also raised a monster, $320m Series H at an $11bn valuation, led by Addition, with Baillie Gifford, QED, and Haun Ventures. (My social post on it is here)
I’m going to be in SF from the 9th to the 30th of August, and I’m looking to line up a handful of founder and C-suite interviews for the Brainfood and tokenized podcasts. HMU if you have something to say.
Looking for the 📣 Weekly Rant? It’s here: 🤖 Who holds AI’s off switch? 👈 - With ChatGPT 5.6 being held back by the US Gov, it looks like it’s not you, or the labs…
Forwarded this email? You can subscribe by hitting the button below. Want to get in touch? Hit reply to this email or the sign up email.
Here's this week's Brainfood in summary
👀 Things to Know:
💸 4 Companies:
Content Corner: Prediction markets price value not settlement
Things to know 👀
It has been quite a week for prediction markets news. First up:
Investigations by The Wall Street Journal and Politico exposed a scandal involving the prediction market platform Polymarket, alleging deceptive advertising and undisclosed promotions. The WSJ says Polymarket created a simulated clone website with a nearly identical URL, replacing a lowercase L with a capitalized I to look like the real domain.
College-aged creators were given access to this fake site to record promotional videos of themselves making wildly successful, simulated bets. In reality, an analysis of these video trades showed that more than half would have lost money in real life, and replicating all of them would have put a user over one hundred and sixty thousand dollars in debt.
Politico uncovered that Polymarket heavily relied on paid progressive and conservative commentators to hype up trading odds without disclosing that they were being sponsored. Over a fourteen-month span, identified influencers posted at least four hundred and ninety times on X without proper advertising disclosures.
🧠What’s most interesting about this story is that close prediction markets are to the media. Because they’re often data providers to media companies, some of the media companies reporting are also ones whose parents have a partnership.
🧠Prediction markets keep being shrouded in controversy, and need a clear run to avoid that. As we see the increased mainstream adoption of these prediction markets and they become more successful, the scrutiny will only intensify.
🧠It’s not all bad news. Insider trading reporting is now increasingly visible. Following the public ase of the former special forces soldier accused of using knowledge to profit from the operation to capture Nicolás Maduro the Venezualan president.
This hasn’t stopped prediction markets from proliferating.
As might competition.
Meta’s Mark Zuckerberg has directed employees to develop a prediction markets app akin to Polymarket and Kalshi, The New York Times reported. The “app would probably rely on a video game-like points system instead” of money, according to one source, though Meta has not ruled out “eventual use of real money betting.” The new app called “Arena” would be separate from Meta’s other platforms like WhatsApp, Instagram and Facebook.
🧠 Is this another distraction like VR for Meta? Meta has spent somewhere in the region of $80bn into VR, which hasn’t returned anything like that.
🧠 Meta has been quite good at copying innovation. Instagram has taken all of Snapchat and TikTok’s best features and monetized them effectively.
🧠 Them doing this with fake currency is obviously avoiding the eye of Sauron from regulators. Meta, more than almost any other company, experienced the ire for governments and regulators when they got kicked for Cambridge Analytica and Libra.
🧠 Whether Meta’s move is a folly or not is not the useful information in this story. The useful information is that prediction markets are winning the war for consumer attention, and that’s something a company of this size should be paying attention to.
They’re certainly large, Volumes at Kalshi and Polymarket, the two largest platforms continue to grow with revenues, especially during the World Cup.
The Commodity Futures Trading Commission (CFTC), which argues it has exclusive jurisdiction over the platforms, brought the case after Kentucky Attorney General Russell Coleman (R) sued Kalshi and Polymarket last week. The agency has been increasing its efforts to assert jurisdiction over prediction markets following CFTC Chair Michael Selig’s confirmation as chair late last year.
The CFTC also recently proposed a comprehensive regulatory framework for prediction markets. These rules would permit broad sports-related event contracts (such as tournament advancements and win-loss results) because they serve price discovery functions. However, the rules would explicitly ban "contrary to public interest" contracts, including bets on player injuries, referee decisions, physical altercations, and discrete "in-game" actions/props.
The framework formally codifies bans on trading contracts tied to illegal or extreme events, such as war, terrorism, and assassinations.
🧠 The regulatory lawfare is in full swing. The problem with clarity by litigation is it doesn’t always work out well for the government. See the last administration’s SEC largely losing to Ripple. When it’s State vs Federal, we’re in uncharted waters.
🧠 And all of this is a distraction from my real worry, which is consumer harm. I’ve seen a lot of proactive work from Kalshi in cracking down on insider trading. If we saw something similar on sensible consumer limits that would be helpful. I get that, if you do that, you look more like gambling than not.
🧠 To their credit, events contracts require significant disclosures today. Prediction markets have disclosures to ensure consumers understand the structure, mechanics, and financial hazards of these investments. There are no guarantees of returns, rules, limits, and the platforms have age restrictions.
🧠 Prediction markets are almost the perfect hedging product. The case for prediction markets, in markets is so much better. The “Will the fed raise rates in 2026” market was pricing a 6% chance in January, 35% on June 22nd, and then 57% by 24th June. What other product could a bank or treasurer buy that would hedge as effectively?
Meta has led a $900 million Series H funding round for Cred and hired the Indian fintech giant's CEO to lead WhatsApp. The funding round values Cred at around $4.5 billion post-money. Cred started as a credit card bill payment app, but has expanded into lending, insurance, wealth, and lifestyle for 17m “credit worthy” Indian’s. Meta will not gain access to any customer information; however, Founder and CEO Kunal Shah replaces Will Cathcart as WhatsApp leader and joins the Meta leadership.
🧠 Was this an invest-hire? Sort of like an acqui-hire but through investing. Cred continues and now has a major backer, but Meta gets what it wants: someone who can help WhatsApp crack India’s payments and finance market.
🧠 WhatsApp is massive in Indian commerce, but nowhere in finance. WhatsApp is a full digital store front, and generates more than $1bn a year from business messaging. Hiring Kunal Shah is a bet they can expand that into payments and business financial services.
🧠 Shah has successfully made large scale financial services “super apps.” If they can build tighter integration, with someone who’s Indian finance native, WhatsApp native, they have a much larger opportunity with over 500m users in the country.
65% of the code Anthropic's product team ships is now written by an AI that lives in their Slack. Now they’ve launched that as a product called Claude Tag. You add @Claude to a channel, point it at the tools and data you pick, and tag it like a coworker. It reads the thread, remembers context across the channel, breaks a task into steps, and posts the work back where the whole team can see it. Flip on ambient mode and it pings you first, surfacing things from across your channels before anyone asks.
One shared Claude per channel. Admins scope what each one can reach, so the sales instance and the engineering instance stay walled off from each other.
🧠 This is the third redesign of how we use LLMs. Andrej Karpathy named three UI patterns. First, a website you visit. Then, an app you download. Now a standing teammate with your org's tools and memory, working next to the humans.
🧠 The UI pattern of tagging agents in Slack is how most advanced users of AI work today. They harden those agents, wire them into tools, and make them able to collaborate across Slack, GitHub, and Linear/Jira. A lab packaging for an enterprise is a logical step.
🧠 I can’t see this outside the context of everyone trying to de-risk their reliance on the big labs. The move towards avoiding vendor lock-in is gaining traction. Microsoft is building the vendor-neutral version through its Agent Framework and the A2A and MCP standards.
🧠 Open source is also coming fast. Paradigm open-sourced Centaur in May. Same tag-it-in-Slack teammate, but it is MIT licensed, self-hosted, and credentials are injected at the network edge, so the agent never touches your raw keys.
🧠 So the question moves past "should we put an agent in chat" to who owns the agent. Rent it from a vendor, and it’s hosted, metered per token, and risks your org's context pooling inside their walls. If you own it, it’s self-hosted, your secrets and your roadmap stay on your side of the firewall.
🧠 Claude Tag is the cleanest version of this I've seen. The open stack sits months behind on polish and well ahead on control. If this becomes the layer where your company's work happens, which side of that trade do you want to be standing on?
Fintech Nerdcon 2026 is coming fast, 19th and 20th November in San Diego. Did you see, we got the Rockstar Games co-founder Jamie King coming. And we’re offering copies of GTA VI for up to 100 ticket purchases. So grab yours now before they’re gone! 👇️
4 Companies 💸
1. Town - The personal productivity AI agent
Town has built an AI agent that helps consumers and business owners get more done. It securely connects to email, documents, and systems to draft documents, and take tasks off your to-do list.
🧠 Are we looking at the next breakout productivity company or Microsoft Clippy? The personal agent for work is still beyond 99% of people. Setting up an OpenClaw is a nightmare; Hermes is better, but still hard. And both are fraught with security issues. Packaging this for SMBs and businesses is solving a real pain. The proof will be in, does it work, or does it have so many early fails, user cohorts drop out before they’ve trained their agent, their workflow? Agents need feedback to get good.
2. Lassie - AI Finance back office for doctors
Lassie claims to run the business side of a doctors practice, handling new enrollments, checking payments, underpayments, claims, and reconciling the paperwork. The company claims to handle 98% of posting automatically. (reconciling payments received from insurance companies).
🧠 Verticalized AI for finance for X is a category. I expect to see more of. Practices are an example of a relatively small business, with complex billing and paperwork that AI (when well-tuned) can be very good at. We had the entire vertical SaaS embedded finance wave, the vertical AI embedded finance category could be much bigger, because instead of helping teams do the work, it does the work. Also, the name. 🐕
3. Vauntly - Agentic Commerce Readiness
Vauntly helps sellers adapt their catalog for LLMs and agents, create new product feeds, and structured data that agents can cleanly find products they need. It gives merchants a score and remedial steps to ensure better conversion. The company says 51% of Gen Z searches start in LLM and its’s a critical channel to get right.
🧠Discovery is the bit of agentic commerce that matters. The big AI labs have all but given up on trying to make people pay inside the chat experience but research is working. A lot of this feels like early e-commerce, where the work is on building catalogs. The problem is, merchants don’t want to be a commodity. They want their brand to matter. So, how much do you want to be discovered if you have a strong brand? And can an LLM represent your brand promise faithfully?
4. Didit - KYC/KYB API for the Agentic Coding era.
Didit provides a single API for KYC, KYB, transaction monitoring and a wide array of compliant onboarding use cases. The product is available via a simple API and claims to support 220 countries, with flat-rate pricing for builders. The team claims to verify millions of individuals per month and offer the fastest verifications on the market.
🧠 Fast verification matters for conversion. The longer a good user has to wait to be onboarded, the more likely they are to churn out before ever using the product. From the looks of things, Didit is an orchestrator platform that sits above lots of regional identity and data providers. This level of scope for a company that just raised a seed is impressive. These guys say they’re profitable already, and in the age of agent-led engineering, it makes sense to have a platform like this that’s native. I just wonder how big these clients get before they want a proper enterprise solution/pricing, and if Didit can scale there.
Good Reads 📚
This fascinating piece by ITO research says prediction markets do something unique. Where a futures or option contract prices the future price of a known thing (e.g. corn), prediction markets price something that today does not have a price at all. They take information that didn’t have a price, like a git repo or academic paper, long before any outcome, and make it a much clearer market signal. That is useful.
🧠 I find so much of the prediction markets conversation is (understandably) dominated by the risk of consumer harm, that we continue to miss that there’s also something novel happening. Both things can be true.
That's all, folks. 👋
Remember, if you're enjoying this content, please do tell all your fintech friends to check it out and hit the subscribe button :)
Want more? I also run the Tokenized podcast and newsletter.
(1) All content and views expressed here are the authors' personal opinions and do not reflect the views of any of their employers or employees.
(2) All companies or assets mentioned by the author in which the author has a personal and/or financial interest are denoted with a *. None of the above constitutes investment advice, and you should seek independent advice before making any investment decisions.
(3) Any companies mentioned are top of mind and used for illustrative purposes only.
(4) A team of researchers has not rigorously fact-checked this. Please don't take it as gospel—strong opinions weakly held
(5) Citations may be missing, and I’ve done my best to cite, but I will always aim to update and correct the live version where possible. If I cited you and got the referencing wrong, please reach out
